Write precompensation optimization in a PRML channel using a selected PRML signal level

Abstract

Write precompensation optimization for a partial response maximum likelihood ("PRML") magnetic recording channel adapted to a mass production environment. Optimal write precompensation may be determined by writing a tribit data pattern known to produce worst case nonlinear transition shift ("NLTS") onto a magnetic recording medium; reading back the recorded data pattern; and calculating the mean-squared error ("MSE") for one or more of the equalized data sample levels associated with the PRML channel to be optimized. The writing and reading back process is repeated with varying amounts of precompensation applied to the written pattern. The optimal amount of write precompensation is determined either by minimizing the MSE for a selected sample level or by minimizing MSE between the middle and the upper or lower of the ternary PRML sample levels.

What is claimed is: 
     
       1. A method of optimizing write precompensation in a PRML magnetic recording channel comprising the steps of: (a) applying a value of write precompensation to a predetermined data pattern thereby creating a precompensated data pattern;   (b) writing said precompensated data pattern onto a magnetic recording medium;   (c) reading said precompensated data pattern from said magnetic recording medium and generating therefrom a read signal;   (d) inputing said read signal into said PRML channel, said PRML channel having an output providing a plurality of PRML signal levels;   (e) sampling said read signal to obtain a series of read signal samples;   (f) selecting one of said plurality of PRML signal levels;   (g) calculating a mean-squared error value of read signal samples from said selected one of said PRML signal levels; and   (h) repeating said steps of applying, writing, reading, inputing, sampling and calculating for said selected PRML signal level using a new value of write precompensation for each repetition until a minimum mean-squared error value of read signal samples from said selected PRML signal level is obtained.   
     
     
       2. The method of claim 1 wherein said PRML magnetic recording channel comprises a class IV partial response channel characterized by the polynomial 1-D 2 . 
     
     
       3. The method of claim 1 wherein said magnetic recording medium is a rigid disk in a disk drive storage device. 
     
     
       4. The method of claim 1 wherein said predetermined data pattern is written around a recording track of a magnetic recording disk in a disk drive storage system. 
     
     
       5. The method of claim 4 wherein said predetermined data pattern comprises a tribit pattern. 
     
     
       6. The method of claim 5 wherein said PRML magnetic recording channel comprises a class IV partial response channel having an,output providing top, middle and bottom signal levels and wherein said selected PRML signal level is said middle signal level. 
     
     
       7. The method of claim 1 wherein said predetermined data pattern comprises a repeating tribit pattern. 
     
     
       8. The method of claim 7 wherein said PRML magnetic recording channel comprises a class IV partial response channel having an output providing top, middle and bottom signal levels and wherein said selected PRML signal level is said middle signal level. 
     
     
       9. The method of claim 1 further comprising the step of storing a value corresponding to said value of write precompensation that yielded said minimum mean-squared error value. 
     
     
       10. The method of claim 9 further comprising the step of using said stored value to select an amount of write precompensation to apply to any data pattern written onto said magnetic recording medium. 
     
     
       11. The method of claim 1 wherein said step of calculating said mean-squared error value comprises the steps of: (a) generating a series of read signal error values by subtracting expected sample values from said read signal samples;   (b) taking the absolute value of said series of error values; and   (c) integrating the absolute value of said series of error values from said selected PRML signal level.   
     
     
       12. The method of claim 1 wherein said step of calculating said mean-squared error value comprises the steps of: (a) generating a series of read signal error values by subtracting expected sample values from said read signal samples;   (b) squaring said series of error values; and   (c) integrating the square of said series of error values from said selected PRML signal level.   
     
     
       13. The method of claim 1 wherein said method is performed in each of a plurality of different frequency recording zones in a zoned recording disk drive storage system and further comprising the steps of saving a minimum mean-squared error value for each zone, selecting a particular zone for a data writing operation, and retrieving and applying a saved mean-squared error value for each selected zone thereby optimizing write precompensation for said selected recording zone before writing data to tracks located within the zone. 
     
     
       14. The method of claim 13 further comprising the steps of storing a series of write precompensation selection values corresponding to the amount of write precompensation that generated minimum mean-squared error during optimization of each of said recording zones; and using one of said stored write precompensation selection values to select an optimal amount of write precompensation when any data pattern is to be written to a selected one of said recording zones. 
     
     
       15. A method of optimizing write precompensation in a PRML magnetic recording channel comprising the steps of: (a) applying a value of write precompensation to a predetermined data pattern thereby creating a precompensated data pattern;   (b) writing said precompensated data pattern onto a magnetic recording medium;   (c) reading said precompensated data pattern from said magnetic recording medium and generating therefrom a read signal;   (d) inputing said read signal into said PRML channel, said PRML channel having an output providing a plurality of PRML signal levels;   (e) sampling said read signal to obtain a series of read signal samples;   (f) selecting a first of said plurality of PRML signal levels;   (g) calculating a mean-squared error value of said read signal samples from said first selected PRML signal level;   (h) selecting a second of said plurality of PRML signal levels;   (i) repeating said steps of reading, inputing, sampling, and calculating to obtain a mean-squared error value of said read signal samples from said second selected PRML signal level; and   (j) repeating said steps of applying, writing, reading, inputing, sampling and calculating a mean-squared error value of said read signal samples from said first selected PRML signal level, and said steps of reading, inputing, sampling and calculating a mean-squared error value of read signal samples from said second selected PRML signal level using a new value of write precompensation for each repetition until a difference between said mean-squared error value of read signal samples from said first selected PRML signal level and said mean-squared error value of read signal samples from said second PRML signal level is minimized.   
     
     
       16. The method of claim 15 wherein said predetermined data pattern comprises a repeating tribit pattern and said magnetic recording medium comprises a magnetic recording disk having a multiplicity of concentrically displaced recording tracks in a disk drive storage system, said repeating tribit pattern written on at least one of said multiplicity of recording tracks. 
     
     
       17. The method of claim 16 wherein said PRML magnetic recording channel comprises a class IV partial response channel having an output providing top, middle and bottom signal levels and wherein said first selected PRML signal level is said middle level and said second selected PRML signal level is either said top or said bottom signal level.